Cyclic di-GMP controls a bacterial cell cycle phosphorylation network

Cyclic di-GMP controls a bacterial cell cycle phosphorylation network

Inauguraldissertation zur

Erlangung der Würde eines Doktors der Philosophie vorgelegt der

Philosophisch-Naturwissenschaftlichen Fakultät der Universität Basel

von

Christoph von Arx aus Olten, Schweiz

Basel, 2018

Originaldokument gespeichert auf dem Dokumentenserver der Universität Basel

edoc.unibas.ch

Genehmigt von der Philosophisch-Naturwissenschaftlichen Fakultät auf Antrag von:

Prof. Dr. Urs Jenal Prof. Dr. Marek Basler

Basel, den 14.11.2017

Prof. Dr. Martin Spiess

Dekan

A BSTRACT

Proliferation by cell division is essential for all living organisms and needs to be tightly controlled. While much is known about metazoan development, cell cycle control in bacteria is less well understood. The bacterial cell cycle is divided into three stages, the period from birth until the initiation of chromosome replication (G1); chromosome replication (S); and the time between the completion of replication and the end of cell division (G2). The lengths of G1 and the G1-to-S phase transition largely determine bacterial proliferation rates as the S and G2 periods remain essentially constant over a wide range of growth rates. Although recent work has improved our understanding of cell growth and cell cycle progression, a molecular frame of the regulatory network driving G1/S transition is still largely missing. Caulobacter crescentus strictly separates its cell cycle stages in an eukaryote-like manner, initiating chromosome replication only once per division cycle. Moreover, Caulobacter divides asymmetrically to generate a replication-competent stalked cell (ST) and a morphologically distinct replication- inert swarmer cell (SW). The latter remains in G1 for a defined period of time before differentiating into a ST cell, a process that is coupled to the initiation of chromosome replication. These unique features make C. crescentus a prime model organism to study bacterial cell cycle control.

Recent findings suggest that C. crescentus cell cycle progression is controlled by oscillating levels of the second messenger c-di-GMP. While c-di-GMP levels are low in SW cells, a sharp c- di-GMP increase directs the SW-to-ST cell transition and mediates S-phase entry when levels of the second messenger peak in stalked cells. In particular, c-di-GMP switches the cell cycle kinase CckA from kinase to phosphatase mode, thereby inactivating the response regulator CtrA, which in its phosphorylated form acts as a replication initiation inhibitor. Here we add another layer of c-di-GMP mediated cell cycle control. We show that c-di-GMP orchestrates the G1/S specific gene expression program by activating the ShkA-TacA phosphorylation pathway. The phosphorylated form of TacA, TacA~P, acts as a transcription factor for several genes involved in the morphogenetic program and in preparing cells for S phase entry. A combination of genetic and biochemical experiments revealed that c-di-GMP controls the ShkA-TacA pathway by directly binding to the ShkA sensor histidine kinase to strongly stimulate its kinase activity. Upon c-di-GMP mediated autophosphorylation, ShkA transfers phosphate to its C-terminal receiver domain from where it is passed on via the phosphotransfer protein ShpA to the transcription factor TacA. We demonstrate that c-di-GMP binds to a pseudo-receiver domain, which is positioned between the kinase core and the

receiver domain and inhibits ShkA kinase activity. Alleviated ShkA auto-inhibition leads to the activation of the ShkA-TacA pathway. We show that c-di-GMP-mediated activation of ShkA and the subsequent c-di-GMP mediated proteolysis of ShkA together define a window of ShkA activity during the cell cycle. The activity of this pathway is further sharpened to a narrow G1/S specific period by TacA degradation, which precedes the degradation of ShkA. Thus, C.

crescentus orchestrates G1/S specific gene expression in high temporal resolution through a combination of c-di-GMP mediated activation and degradation of ShkA-TacA phosphorylation components. Because ShkA binds c-di-GMP with higher affinity than CckA, the ShkA-mediated G1/S transcriptional network precedes the CckA switch leading to chromosome replication initiation as c-di-GMP levels are building up during the cell cycle. Together, these results propose a model, where an upshift of c-di-GMP drives consecutive steps of the G1/S cell cycle progression in C. crescentus by sequentially activating hierarchized phosphorylation pathways.

C ONTENTS

1 I NTRODUCTION ______________________________________ 1

1.1 Regulation of cell cycle progression ________________________________________ 1 1.2 Systems to transfer information across the membrane ________________________ 2 1.3 Proteolysis ___________________________________________________________ 9 1.4 Second messengers in bacteria __________________________________________ 10 1.5 Alphaproteobacteria __________________________________________________ 13 1.6 Caulobacter crescentus ________________________________________________ 14 1.7 Caulobacter crescentus cell division ______________________________________ 16

2 A IM OF THE T HESIS __________________________________ 22 3 P ^ROJECT A ^BSTRACT __________________________________ 23 4 P ROJECT I NTRODUCTION ______________________________ 24 5 R ^ESULTS _________________________________________ 28

5.1 C-di-GMP controls the ShkA-TacA phosphorelay_____________________________ 29 5.2 C-di-GMP stimulates ShkA kinase activity __________________________________ 33 5.3 C-di-GMP binding alleviates ShkA auto-inhibition ____________________________ 36 5.4 C-di-GMP binds to the ShkA REC1 pseudo-receiver domain ____________________ 40 5.5 C-di-GMP-mediated activation and proteolysis defines the ShkA-TacA activity window during the cell cycle _______________________________________________________ 45 5.6 The ShkA-TacA pathway limits gene expression to G1/S _______________________ 48

6 D ^ISCUSSION & O ^UTLOOK ______________________________ 53 7 M ATERIAL & M ETHODS _______________________________ 59

7.1 Strains used in this study _______________________________________________ 59 7.2 Plasmids ____________________________________________________________ 61

7.3 Growth conditions ____________________________________________________ 69 7.4 Phos-Tag immunoblotting ______________________________________________ 70 7.5 Immunoblotting ______________________________________________________ 70 7.6 Capture Compound Mass Spectrometry (CCMS) _____________________________ 70 7.7 Protein expression and purification _______________________________________ 71 7.8 Kinase in vitro assays __________________________________________________ 71 7.9 Isothermal titration calorimetry (ITC) measurements _________________________ 71 7.10 β-galactosidase measurements __________________________________________ 72 7.11 Microscopy __________________________________________________________ 72 7.12 Selection/screen for c-di-GMP independent mutations in shkA _________________ 72

8 R EFERENCES ______________________________________ 74

9 A CKNOWLEDGMENTS ________________________________ 87

10 C URRICULUM V ITAE _________________________________ 88

1 I NTRODUCTION

Bacteria are omnipresent in our environment and can survive and grow almost everywhere, even in extreme environments¹. To live in different niches, bacteria have evolved a high variety of specialized enzymes, that can even be used in biotechnological applications². Some industries rely heavily on bacteria, which are important for the treatment of wastewater³ or can be used to produce antibiotics. Bacteria can also be utilized for the production of amino acid on a million-ton scale, for example to produce the flavor enhancer Glutamic acid⁴. Traditionally, bacteria are an important component in the production of foods like yoghurt, cheese, sourdough, sauerkraut, or kimchi.

Colonizing almost every ecosystem, bacteria are heavily involved in forming the biosphere of our planet. In the holobiont, the ecological system formed between plants and their microbiome, plants and bacteria influence each other and are interdependent⁵. On the other hand, bacteria can have a negative effect on plant growth, for example by leading to losses in food harvest⁶. Likewise, bacteria reside in and on the bodies of animals and humans. Bacteria can live in symbiosis with the animal or human hosts, affecting the metabolism and immune system positively⁷. Conversely, pathobionts invading the host’s microbiota can lead to infections.

Because bacteria influence humans and their environment in so many ways, deepening our understanding of bacterial growth and behavior is essential for our well-being. Among others, this knowledge will help us to find treatments for infections, improve crop harvest, optimize industrial processes, and possibly solve the energy needs of our industrialized societies.

1.1 R

Proliferation by cell division is essential for all living organisms and needs to be tightly controlled to provide appropriate numbers of specialized cells and to avoid pathological side effects like cancer. In metazoan tissue, stem cells need to proliferate and specialize.

Coordinating cell division and cell differentiation is thereby central⁸. In eukaryotic cells, a highly complex network of cyclin-dependent kinases and protein degradation drives the process of the unidirectional cell cycle transition. Different cyclins oscillate to time the different stages in eukaryotic cell division. This process begins with the synthesis of specific cyclins at the beginning of the next cell cycle phase. These cyclins then activate cyclin-

2 | Introduction

dependent kinases (Cdks) by binding to them. The so formed cyclin-Cdks complexes regulate specific events to progress the cell cycle like chromatin remodeling or centrosome duplication.

Degradation then removes the current cyclin and thereby deactivates the Cdks. This allows the next cyclin to take over⁹.

Conceptually, the aquatic bacterium Caulobacter crescentus controls cell cycle progression similarly to eukaryotes. In C. crescentus, three regulatory mechanisms drive the unidirectional cell cycle progression: A complex network of two-component systems advances cell division, and proteolysis specifically removes proteins that are not needed anymore. The third regulatory mechanism consists of the oscillating bacterial second messenger c-di-GMP to orchestrate the timing of cell cycle progression¹⁰. The following chapters will introduce each of these mechanisms in detail.

1.2 S

1.2.1 One-component systems

The most widely used mechanism for prokaryotes to sense their environment is by utilizing one-component signaling systems. These systems consist of proteins with two specialized domains: An external input domain that senses a specific environmental signal, and an internal output domain that leads to a cellular response (Figure 1.1). One-component systems can sit in the membrane and sense environmental stimuli. The corresponding output is often carried out by a HTH domain that can bind to a ligand or DNA. For example, ToxR is a transmembrane transcription factor regulating virulence in Vibrio cholerae. ToxR consists of a cytosolic DNA binding domain and a periplasmic domain that senses pH and bile salts¹¹. In a similar way, CadC in Escherichia coli also senses pH and induces the transcription of stress response genes¹². One-component systems are encoded in the genomes of almost all bacteria and archaea¹³. This is exemplified by the archaeon Sulfolobus acidocaldarius that regulates swimming with the one-component system ArnR¹⁴.

Proteins with a similar architecture can also remain in the cytosol where they bind environmental cues that can penetrate the cell envelope or are imported. One example for a cytosolic one-component regulator is the tetrameric E. coli protein LacI. LacI binds to the promoter of the lac operon and inhibits transcription by forming a DNA loop¹⁵. In its DNA bound form, the LacI tetramer is stabilized by interactions between the regulatory domains.

Binding of the LacI substrate allolactose or IPTG induces a hinge-like motion within the protein,

releasing this interaction. This leads to a conformational change in the DNA-binding domain, releasing LacI from the DNA¹⁶.

One-component systems that reside in the membrane are limited in binding to DNA, while cytoplasmic one-component systems are restricted in sensing the outside. This limitation likely led to the evolution of two-component systems. In those systems, the input and the output domain are separated, and phosphorylation links the two proteins together. This separation allows the input domain to sit in the membrane, interacting with signals from the outside, while the output domain remains in the cytoplasm and can directly interact with DNA¹⁷. 1.2.2 Two-component systems and phosphorelays

Similar to one-component systems, two-component systems allow bacteria to respond to their environment. These systems fulfill many roles in cellular signaling circuits. The most basic set- up consists of two proteins, the histidine kinase (HK) and the response regulator (RR). A sensor HK usually has an input domain and always two kinase domains, the catalytically active CA domain and the DHp domain with a conserved His residue that can be phosphorylated. The RR contains a highly conserved receiver domain with an Asp residue onto which the phosphate is transferred. This leads to a conformational change and an output (see below and Figure 1.1).

Sensor histidine kinases can react to a wide variety of inputs, from light, temperature or gas to specific low molecular weight molecules like amino acids or nucleotides¹⁸. Upon sensing a signal, the HK autophosphorylates on the conserved His residue. Subsequently, the phosphate is then transferred to an Asp residue on the RR, which in turn activates an effector domain to carry out a task¹⁹. The basic two-component system can be adapted to a more complex phosphorelay (Figure 1.1). In this case the phosphorylation scheme is HisAspHisAsp over multiple proteins. Phosphorelays often involve hybrid kinases in which the histidine kinase and the first receiver domain are fused together, and phosphotransfer proteins that shuttle the signal to the response regulator²⁰. These more complex systems can be used to branch a signal to multiple response regulators (one-to-many), for example in the E. coli chemotaxis pathway, or to integrate signals from multiple kinases on one response regulator (many-to-one)²¹. One example for a one-to-many pathway can be found in the central C. crescentus CckA-ChpT-CtrA pathway (see below). Here the phosphotransfer protein ChpT donates phosphoryl groups to the response regulators CtrA and CpdR, which act as transcription factor and as protease adaptor, respectively²².

The function of a response regulator is ultimately determined by the biochemical activity of its output domain. These are diverse but consist mostly (63 %) of DNA binding domains²³. These

4 | Introduction

domains include the OmpR-like winged-helix domain, the NarL four-helix helix-turn-helix domain, the NtrC-like helix-turn-helix domain and the LytR-like DNA binding domain with an unusual β-fold²³. Alternatively, these domains can mediate protein-protein interactions²⁴ or have an enzymatic output. For instance they can produce second messengers like bis-(3' 5')- cyclic diguanylic acid (c-di-GMP) with a dedicated GGDEF domain²⁵. Members of this protein family lacking an output domain are called single domain response regulators and can have multiple cellular functions. A prominent example is the E. coli chemotaxis protein CheY that mediates motility control via its interaction with the flagellar motor switch²⁶ (Chapter Chemotaxis, Page 7) or DivK from C. crescentus, which in its phosphorylated form allosterically controls the activity of a histidine kinase, thereby interconnecting two phosphorylation pathways²⁷ (see below).

Figure 1.1: Schematic diagrams of a one-component system, a two-component system and a phosphorelay. (A) The most basic signaling system comprises of a single protein with an input domain and an output domain. (B) In general, two-component systems consist of a histidine kinase and a response regulator. The CA domain of the histidine kinase binds ATP with the highly conserved N, G1, F, and G2 box. The phosphate is transferred to the His residue of the DHp domain. The phosphate is then transferred to the Asp residue on the receiver domain of the response regulator. (C) The more complex phosphorelay extends on the two-component system in a highly modular fashion. With an additional receiver domain that is often fused to the histidine kinase, and a histidine phosphotransferase, the phosphorylation scheme is His-Asp-His-Asp over multiple proteins. Adapted from^13,23.

Many sensor histidine kinases are bifunctional enzymes that can also act as phosphatases^28,29. This allows the pathway to suppress nonspecific phosphorylation of its own response regulator. Additionally, by switching a sensor histidine kinase from its kinase to its phosphatase mode, the cellular function controlled by this pathway can be quickly and efficiently shut off³⁰.

Two-component systems are widely distributed in bacteria and are also found in all other domains of life. Their modular architecture makes histidine kinases and response regulators versatile signaling pathways, connecting a large variety of upstream signals with different downstream processes. However, even though input and output mechanisms are different, structure and function of the central domains involved in phosphorylation are well conserved²³. Histidine kinases form homodimers and two domains define the cytoplasmic kinase core. The CA domain is catalytically active and binds ATP. Structurally it consists of a five-stranded β-sheet and three α-helices packed against it that form an α/β-sandwich³¹. This domain contains specific motifs, the N, G1, F, and G2 box, that form a highly conserved ATP binding cavity³². The phosphate is then transferred to the conserved His residue that resides in the DHp domain (Figure 1.1). This central kinase domain is responsible for dimerization, phosphotransfer and phosphatase activity. The DHp consists of two α-helices that form a four- helix bundle with the other protomer^33,34. A conformational change between the CA and the DHp domain, mediated by the input domain, regulates autophosphorylation by moving the CA domain towards the DHp domain³⁵. The conformation of the loop at the base of the helical DHp dimerization domain determines whether the histidine kinase autophosphorylates in trans or in cis³⁶.The catalytic mechanism of autophosphorylation is independent of cis- or trans-directionality, however it might be important for response regulator specificity³⁷. The process of autophosphorylation and phosphotransfer is separated, as phosphotransfer requires the movement of the CA domain to free the access to the phosphorylated His residue for the receiver domain. However, these two reactions still occur simultaneously in neighboring protomers of a dimer. Each protomer then switches to take over the function of the other protomer with an conformational change, a process that might be triggered by the interacting receiver domain³⁸. To control the activation of the pathway, two-component systems often display phosphatase activity. The phosphate travels back from the response regulator and is removed by the DHp domain of the histidine kinase. However, this is not just a reversal of kinase activity, as the His residue of the DHp domain is not phosphorylated in this

6 | Introduction

process³⁹. Nonetheless, it was shown that conserved key residues in the DHp domain are necessary for phosphatase activity⁴⁰.

The abundance of two-component systems in many bacteria requires high specificity to avoid crosstalk between different signaling pathways. A small subset of coevolving residues in the interface of histidine kinases and response regulators is enough to determine specificity⁴¹. These amino acids can be artificially exchanged to change specificity of a histidine kinase to another response regulator⁴². Response regulators themselves contain highly conserved receiver domains, formed by a central five-stranded parallel β-sheet and five α-helices surrounding it⁴³. Receiver domains are not passive, they catalyze their own phosphorylation and dephosphorylation reactions. For this reaction five essential residues are necessary: A divalent metal ion required for phosphorylation is bound by three Asp residues, while Lys and Thr/Ser residues are crucial for signal transduction⁴⁴. Phosphorylation leads to a conformational change of the distal α4-β5-α5 surface, which in turn mediates interaction with the neighboring output domain protomer or with other proteins⁴⁵.

Examples of receiver domains without the key residues for phosphorylation can be found in multiple species^46,47. Pseudo-receiver domains can act as sensory domains. CikA, a protein of Synechococcus elongates circadian system contains a pseudo receiver domain that binds a small organic quinone molecule. Upon binding of its ligand, the protein becomes quickly degraded. This allows the cell to measure light indirectly over the cells redox state⁴⁸. Similarly, it was shown recently that a subclass of CheY-like single domain response regulators interact with the flagellar motor not in response to phosphorylation, but in response to binding c-di- GMP (Nesper et al. 2017, in revision).

In the more complex phosphorelays, as represented for example by the CckA-ChpT-CtrA pathway in C. crescentus, a histidine phosphotransferase shuttles the phosphate from the (hybrid) histidine kinase to the response regulator (Figure 1.1). Two different types of phosphotransferases can be found in bacteria: dimeric transferases like ChpT and monomeric transferases like ShpA from the C. Crescentus ShkA-TacA pathway. ChpT structurally represents a histidine kinase, the CA domain however is not functional. The site of phosphorylation resembles a dimeric four-helix bundle similar to a DHp domain⁴⁹. The monomeric ShpA also forms a four-helix bundle, but different to ChpT it has no degenerate CA domains⁵⁰.

1.2.2.1 Chemotaxis

The best understood example of bacterial signal transduction is chemotaxis. Many free- swimming bacteria and archaea move towards attractants, or away from repellents, by sensing molecule concentrations. E. coli perform a biased random walk by measuring changes of gradients while swimming in one direction. Worsening conditions or no gradient change leads to earlier tumbling and directional change. This means bacteria can sense current ligand concentrations and remember past ligand concentrations at the same time⁵¹. Transmembrane chemoreceptors, called methyl-accepting chemotaxis proteins (MCPs), mediate this behavior.

MCPs form a complex together with the histidine autokinase CheA and its regulator CheW.

Phosphorylation activity of CheA is dependent on ligand binding and receptor sensitivity, which in turn is regulated by methylation of specific modification sites. Methylation of these sites decreases receptor sensitivity to its ligand, likely by inducing conformational changes⁵². Without ligand, CheA is active and phosphorylates the response regulator CheY, which interacts with the flagellar motor to induce tumbling. After ligand binding of the MCP, the kinase is turned off, leading to dephosphorylation of CheY by the phosphatase CheZ and subsequently longer swimming phases. In this state, the methyltransferase CheR decreases sensitivity of the MCP to its ligand by methylating specific sites in the dimer. The subsequent loss of ligand binding means that a steady or decreasing ligand concentration activates CheA.

As a consequence, CheA phosphorylates and thereby activates the methylesterase CheB. The removal of methyl-residues increases MCP sensitivity once again. In addition, CheA activity also induces tumbling, allowing the bacterium to change direction. This feedback loop leads to a memory of a few seconds⁵³ (Figure 1.2).

Although MCPs in E. coli are the most studied and best understood, this unique prokaryotic chemotaxis system is highly conserved in bacteria genomes⁵⁴. In C. crescentus, chemoreceptors are expressed only in the swarmer cells and localized to the flagellar pole where they are partially ordered in a complex array, similar to other bacteria⁵⁵. E. Coli MCPs expressed in C. crescentus are localized to the same pole, further emphasizing the highly conserved nature of these receptors⁵⁶. Initial binding of a ligand leads to a small conformational change within the receptor dimer. This conformational change is then transmitted through the entire array, greatly amplifying the initial signal⁵⁷.

Next to many repellents and attractants like sugars and other nutrients, bacterial chemotaxis can also react to AI-2, a general signal of quorum sensing⁵⁸.

8 | Introduction

Figure 1.2: Chemotaxis pathway: Methyl-accepting chemotaxis proteins (MCPs) form a complex with the histidine autokinase CheA and its regulator CheW. CheA is only active when no ligand is bound to the MCP and its phosphorylation activity controls two separate modules: (1) Motor control and (2) sensory adaption. (1) Motor control is achieved by the phosphorylation of the response regulator CheY. CheY switches the flagellar motor from a CCW rotation to CW rotation, leading to tumbling. When no ligand is bound and CheA is inactive, the phosphatase CheZ inactivates CheY, leading to a longer swimming phase. (2) Sensory adaption allows the bacterium to respond to ligand gradients. Without a ligand bound, CheA phosphorylates the methylesterase CheB, thereby activating it.

Removal of methyl-residues from the MCP increases its sensibility until it can bind a ligand. This deactivates CheA and allows CheR to methylate the MCP to decrease sensibility of the ligand. Together, these two modules allow the adaption to molecule gradients. If a ligand concentration remains steady or decreases, MCP sensibility will decrease and induce tumbling, while at the same time increasing MCP sensibility again. Adapted from⁵⁷.

1.2.2.2 Chemoreceptors with other functions

Apart from the mediation of chemotaxis, chemoreceptors can also have other functions, many of them unknown. Chemoreceptors are structurally diverse and frequently found in bacteria⁵⁹. In Pseudomonas aeruginosa, the Wsp signal transduction complex is homologous to chemoreceptor systems mediating chemotaxis. However, in this case activation of the chemoreceptor WspA leads to an increase of biofilm formation⁶⁰. The homologue of CheY in this system, WspR, has an GGDEF-domain that produces c-di-GMP. WspA is very similar to other MCPs, but it is not localized to the cell pole due to few amino acid changes⁶¹. Another example in the same bacterium is the Chp system. Upon receiving its signal, it modulates cAMP levels and twitching motility⁶².

1.3 P

For all domains of life, it is a necessity to remove proteins in a controlled fashion. Protein homeostasis must be upheld by the degradation of damaged proteins or protein fragments from mistranslation or cleavage. Just as important is protein degradation to control cellular processes like cell-cycle progression or differentiation, and degradation as a stress response.

Responsible for the active removal of proteins are oligomeric AAA+ proteases⁶³. Numerous AAA+ protease complexes exist in bacteria with the basic architecture of an AAA+ ring that denatures proteins, and a compartmental protease degrading the proteins⁶⁴. In the paradigmatic AAA+ protease ClpXP complex, the chaperone ClpX targets specific proteins to deliver them to the protease ClpP. ClpXP can thereby even degrade aggregated proteins that form under stress conditions⁶⁵. In C. crescentus, ClpXP is essential. This is due to the toxin SocB that accumulates and leads to cell death without degradation by ClpXP. If SocB is lacking, ClpX can be deleted, however affected cells show deficits in growth and morphology⁶⁶.

ClpX recognizes proteins by short and unstructured peptide sequences. One example is the SsrA-tag in E. coli that is recognized by both ClpXP and ClpAP. The SsrA tag is added to proteins when translation is stalling and allows the cell to quickly remove these potentially harmful protein fragments⁶⁷. Another example is the phage protein MuA, which is also recognized by ClpXP⁶⁸. In total, ClpXP binds to at least five different degradation motifs, three N-terminal and two C-terminal. The C-terminal motifs share either homology with the hydrophobic SsrA recognition motif (LAA-COOH) or the MuA recognition motif (RRKKAI-COOH)⁶⁹. Two proteins that are targeted by ClpXP with an SsrA homology motif are the C. crescentus proteins CtrA (NAA-COOH) and TacA (EAG-COOH). This specificity can be negated with a double mutation of the C-terminus to the negatively charged amino acid aspartic acid, for example in TacA with EAG-COOH to EDD-COOH⁷⁰.

The SsrA tag requires another protein, SsrB, to be efficiently targeted to its protease. SsrB thereby acts as an adaptor of SsrA⁷¹. Adapter proteins expand and diversify the substrate recognition capabilities of proteases and allow specific cell responses to altering conditions⁷². In the Gram-positive bacterium Bacillus subtilis, CtsR is a repressor of clp gene expression and therefore proteolysis. CtsR is active at a basal level under normal conditions. When exposed to stress, the protein McsA was found to activate the adapter protein McsB. McsB targets CtsR to the protease ClpCP where it is degraded. Subsequently, this allows the upregulation of stress response genes⁷³. Adaptor proteins can also modulate protease specificity directly. The

10 | Introduction

adaptor ClpS binds to the protease ClpAP and inhibits degradation of SsrA-tagged proteins. At the same time, ClpS enhances degradation of protein aggregates⁷⁴.

Protein degradation is often part of cell cycle regulation and then needs to be exactly timed and regulated. This additional level of control can also be achieved with adaptor proteins, allowing the degradation of a specific target at a specific time⁷⁵. The adaptors themselves are also degraded by ClpXP, however they are shielded from degradation by binding their specific substrate⁷⁶.

1.4 S

1.4.1 The role of c-di-GMP

C-di-GMP was first described as a regulator of cellulose synthesis in Acetobacter xylinum⁷⁷, and is now recognized as a major second messenger that coordinates major decisions in bacterial life style and in cell cycle progression⁷⁸. In general, cells low on c-di-GMP are motile, while a c- di-GMP increase switches the cell to a sessile lifestyle in multicellular biofilms (Figure 1.3). In addition, c-di-GMP is thought to be involved in regulating virulence. While motile low-c-di- GMP bacteria are responsible for acute infections, high-c-di-GMP cells are involved in chronic infections⁷⁹.

The second messenger c-di-GMP itself consists of two guanine bases that are linked to a cycle by ribose and phosphate⁸⁰. It was suggested that c-di-GMP binds to its target likely as a dimer⁸¹. However, under physiological conditions c-di-GMP is most likely available as a monomer⁸². Within the bacterial cell, c-di-GMP levels and the complex, corresponding responses are regulated by the antagonistic action of diguanylate cyclases that produce c-di- GMP, and phosphodiesterases that degrade c-di-GMP^77,83,84 (Figure 1.3). These proteins are widespread and abundant in bacterial genomes⁸⁵.

The asymmetric cell cycle of C. crescentus is strongly dependent on c-di-GMP with the swarmer cell having low levels and the predivisional and stalked cell having higher levels of c- di-GMP⁸⁶. A mutant strain lacking all diguanylate cyclases (cdG⁰) loses cell polarity and cell differentiation, expresses no stalks, pili, holdfast, or flagellum, resulting in a phenotype of elongated and apolar cells⁸⁷.

While c-di-GMP is abundant in the phylum Proteobacteria, c-di-GMP is absent in other phyla like Bacteroidetes or Fusobacteria⁸⁸. However, these bacteria could potentially be controlled by other second messengers like c-di-AMP. Interestingly, in some bacteria of the phyla

Firmicutes and Actinobacteria, both signaling systems, c-di-AMP and c-di-GMP, might be present⁸⁹.

1.4.2 Diguanylate cyclases and phosphodiesterases

Diguanylate cyclases contain a GGDEF domain which is necessary for c-di-GMP production⁹⁰. GGDEF domains are highly conserved over bacterial kingdoms and the domain itself is sufficient for c-di-GMP production, however regulatory domains are necessary for high activity⁹¹. GGDEF domains act in a similar way as adenylate cyclases, however with a different nucleotide-binding mode. For c-di-GMP synthesis, two GGDEF domains are necessary, requiring dimerization of the diguanylate cyclases⁹². For c-di-GMP formation, it is likely that two GTP molecules have to be arranged in a antiparallel orientation by the GGDEF domains⁹³. Different mechanisms have evolved to achieve this interaction between GGDEF domains that provide a regulatory basis for c-di-GMP synthesis⁹⁴. Additional control of c-di-GMP production is provided by allosteric inhibition of some GGDEF domains by c-di-GMP itself⁹⁵.

As direct antagonists to diguanylate cyclases, phosphodiesterases hydrolyze c-di-GMP. One major class of phosphodiesterases contains an EAL domain. This domain hydrolyzes c-di-GMP into linear dimeric GMP (pGpG) in an Mg²⁺ or Mn²⁺ dependent manner. In contrast, Ca²⁺ can inhibit phosphodiesterase activity⁹⁶. The activity of some phosphodiesterases is dependent on allosteric activation by GTP. In this case, inactive GGDEF domains can regulate EAL domains by binding the required GTP⁹⁷. Activity of EAL domains can be regulated by conformational changes induced by an input domain that catalytically optimizes the active site. This is the case in the light-sensing protein BlrP1 in Klebsiella pneumoniae⁹⁸. The product of phosphodiesterases with EAL domains pGpG needs to be further hydrolyzed. In P. aeruginosa, the oligoribonuclease Orn can degrade pGpG to two GMPs and thereby remove pGpG from the cell⁹⁹.

Another widely distributed type of phosphodiesterases utilizes HD-GYP domains to directly convert c-di-GMP to two GMPs¹⁰⁰. However, a first hydrolysis step produces pGpG. HD-GYP domains can also accept pGpG as a primary substrate and further hydrolyze it to two GMPs^101,102.

12 | Introduction

Figure 1.3: Diguanylate cyclases containing a GGDEF domain produce c-di-GMP from two GTP molecules. Removal of c-di-GMP is carried out by phosphodiesterase with either an EAL domain, producing linear dimeric GMP (pGpG), or an HD-GYP domain that hydrolyzes c-di-GMP to two GMP in a two-step process. High cellular c-di-GMP levels are generally associated with a decreased motility, biofilm formation and modulated virulence.

1.4.3 C-di-GMP receptors

A broad variety of c-di-GMP receptors can be found in bacteria. A strong and common c-di- GMP receptor is resembled by PilZ domain proteins that can be found in various bacterial species⁸⁴. PilZ domains can be dimeric or monomeric and express a wide-range of mechanistic principles and binding stochiometries. However, in general they lead to a allosteric rearrangement with a hinge-like movement, after c-di-GMP binds to the PilZ signature consensus RxxxR¹⁰³.

GGDEF domains can also bind c-di-GMP and act as receptors⁹⁵. An example of a c-di-GMP receptor with a degenerated GGDEF motif, which is catalytically inactive, is the C. crescentus adaptor-protein PopA¹⁰⁴. In a similar way, degenerate phosphodiesterases can act as a c-di- GMP receptors with an inactive EAL domain. This is the case for FimX, a P. aeruginosa protein¹⁰⁵.

It is important to note, that many c-di-GMP receptors do not show a common motif or a conserved domain. An example without a general c-di-GMP-binding domain can be found in P.

aeruginosa, where the transcriptional regulator FleQ is de-repressed upon binding to c-di- GMP¹⁰⁶. In C. crescentus, the histidine kinase CckA binds c-di-GMP on its CA domain. This leads to an interaction with the DHp domain that allows the kinase to switch from a kinase to a phosphatase state²⁹.

Not only proteins can bind c-di-GMP. A special class of c-di-GMP binders is represented by RNA molecules. C-di-GMP binding bacterial riboswitches are highly specific and can undergo a global structural rearrangement upon ligand binding¹⁰⁷. Riboswitches can have multiple functions and c-di-GMP binding can regulate transcriptional regulation or translation¹⁰⁸. One example is the Vibrio cholerae class I riboswitch Vc1 that, in response to c-di-GMP, positively regulates downstream gene expression of a colonization factor¹⁰⁹.

1.4.4 Alternative second messengers in bacteria

Apart from c-di-GMP, many more linear and cyclic nucleotides are known to act as second messengers in bacteria. These signals include cAMP¹¹⁰, cGMP¹¹¹, (p)ppGpp¹¹², c-di-AMP¹¹³ and even hybrid cyclic AMP-GMP molecules¹¹⁴ that are involved in pathogenicity of bacteria¹¹⁵. The second messenger cAMP was first described as an alarmone in in E. coli, that responds to glucose levels¹¹⁶. In low-glucose conditions, cAMP is produced by an adenylate cyclase and regulates biofilm formation positively¹¹⁷. However, cAMP is recognized as an essential signal involved in bacterial virulence in several pathogens by now¹¹⁸.

The linear nucleotide (p)ppGpp is another alarmone that is synthesized under stress conditions or during starvation. Thereby, (p)ppGpp can have a multitude of effects on bacteria including development and virulence¹¹⁹. In C. crescentus, the enzyme SpoT synthesizes (p)ppGpp under starving conditions. Increasing levels of the (p)ppGpp alarmone influence levels of the proteins DnaA and CtrA directly, inhibiting G1-to-S phase transition and keeping the cell in the swarming state¹²⁰.

Several Gram-positive bacteria where shown to use the second messenger c-di-AMP. There, c- di-AMP can control essential processes like cell wall homeostasis or ensures DNA integrity⁸⁹. In Bacillus subtilis and related pathogens, c-di-AMP is essential and controls potassium homeostasis¹²¹.

As many of those molecules were only discovered recently and considering the abundance of GGDEF domains in bacterial genomes¹²², it is likely that even more roles of second messenger signaling in bacteria will emerge in the future.

1.5 A

Alphaproteobacteria is an omnipresent class of gram-negative bacteria with diverse ecological niches, from free-living environmental species to host-associated pathogens¹²³. For example, the intracellular pathogen Rickettsia prowazekii that is responsible for epidemic typhus in

14 | Introduction

humans¹²⁴, is also thought to be a close ancestor to mitochondria¹²⁵. Alphaproteobacteria can have positive or negative effects on plants. The plant family Leguminosae can form a specific symbiose with many different Rhizobiales, for example Rhizobium, to fix atmospheric nitrogen¹²⁶. However, plants can be affected by the pathogen Agrobacterium tumefaciens that can ontogenically transform plant cells¹²⁷. Some Alphabroteobacteria can grow under extreme conditions, like Polymorphobacter multimanifer that lives within stones in Antartica¹²⁸.

Even though there is a high diversity in Alphaproteobacteria, the cell cycles of bacteria in this class still rely on similar mechanisms with conserved central pathways¹²⁹. The oligotrophic bacterium C. crescentus is a bacterial model organism with an obligate asymmetric cell division that gives rise to two distinct cell types, a motile replication-inert swarmer cell and a sessile replication-competent cell^130,131.

1.6 C

The ubiquitous Caulobacter genus was initially defined by the ability to form stalks¹³². Stalks are cell envelope extensions that are always polar in Caulobacter, however can also be subpolar or bipolar in another Alphaproteobacterium genus, Asticcacaulis¹³³. The stalk of Hyphomonas neptunium, also belonging to the same genus, is directly involved in cell division.

This bacterium uses a budding mechanism, in which the daughter cell grows at the end of the stalk. In this asymmetric process, the emerging cell does not have a stalk¹³⁴. In C. crescentus, stalks are the hallmark of the replication-competent cell type. The cytoplasmic composition of the stalk lacks most proteins of the cell body, however it includes proteins for nutritional uptake¹³⁵. In addition, the long and thin form of the stalk is ideal for the uptake of diffusing nutrients¹³⁶. The end of the stalk consists of a strong adhesive holdfast, allowing the bacterium to attach to surfaces^137,138.

After cell division, genome replication is inhibited in the motile swarmer cell (G1 phase).

During the irreversible transition into a sessile stalked cell (S phase), the cell starts the replication cycle. Swarming daughter cells are then shed off from the opposite pole of the stalk¹³⁹ (Figure 1.4). If the new-born swarmer cell is in contact with a surface, a holdfast is immediately formed¹⁴⁰. The characteristic curved (crescent) shape of C. crescentus helps the predivisional cell to contact the surface, enhancing colony formation¹⁴¹.

Figure 1.4: The C. crescentus cell cycle: Motile swarmer cells differentiate to sessile stalked cells by losing the flagellum and pili, replacing them with a holdfast and later the stalk. The stalked cell begins replication of the genome, initializing an asymmetric cell division. The stalked cell remains replication-competent, while the daughter cell differentiates into a replication-inert swarmer cell. The orange color indicates rising c-di-GMP levels during G1- to-S phase transition, while the swarmer cell does not have c-di-GMP (blue). The peak of c-di-GMP levels is qualitatively shown as a grey curve.

16 | Introduction

1.7 C

To undergo the complex process of cell differentiation and division, C. crescentus uses an intricate set of mechanisms that is tightly controlled. Protein expression, levels of c-di-GMP and protein degradation together irreversibly allow the cell to transit through G1-to-S-phase.

The following chapters describe the C. crescentus cell division in detail.

1.7.1 The initiation of chromosome replication

A central step of cell division is the controlled replication of the chromosome. In bacteria, a single origin of replication (oriC) acts as starting site for chromosome replication. This genetic region likely encodes species-specific instructions for the binding and regulation of proteins that form the initiation complex. Core of the initiation complex is the initiator protein DnaA.

The dnaA gene is present in almost all bacteria with the exception of few obligate endosymbiotic species¹⁴².

As an AAA+ type protein, DnaA binds ATP with high affinity and after reaching peak level, forms multimeric complexes on the oriC region. Mediated by this complex and other proteins, duplex DNA can unwind and the DNA helicase is allowed to initiate the construction of replication forks¹⁴³. In fast growing bacteria like E. coli, another round of replication can be initiated before duplication has finished, dramatically speeding up cell division time. C.

crescentus on the other hand is limited to one replication cycle at the time¹⁴⁴.

DnaA is controlled on many levels, including expression, ATP hydrolysis and ultimately protein degradation¹⁴⁵. However, C. crescentus adds an additional layer of control. While DnaA oscillates in the stalked cell to constantly replicate the chromosome, DnaA is inhibited in the swarmer cell by the response regulator CtrA. There, CtrA in its phosphorylated state binds to the oriC and thereby silences the origin of replication, ensuring the asymmetric life cycle of C.

crescentus. For replication to start, CtrA has to be dephosphorylated and subsequently removed from the cell¹⁴⁶.

1.7.2 The master cell fate regulators of Caulobacter crescentus control the CckA-ChpT-CtrA pathway and c-di-GMP levels during the G1-to-S phase transition

Central in C. crescentus cell division is the activity and the localization of the kinases DivJ and PleC, which control the master cell fate regulator DivK in a complex phosphorylation network¹⁴⁷. These factors together control the activity of the CckA-ChpT-CtrA pathway and modulate c-di-GMP levels within the cell, which in turn together influence protein degradation¹⁴⁸.

The central transcription factor CtrA is essential for C. crescentus cell division and development¹⁴⁹. In addition to controlling almost hundred cell cycle-regulated genes directly¹⁵⁰, including TacA¹⁵¹, CtrA also specifically represses DNA replication in the swarmer cell by binding to the chromosome replication origin of C. crescentus. Therefore, to enter S- phase, CtrA first has to be dephosphorylated and then removed from the swarmer cell during G1-to-S phase transition¹⁵².

CtrA itself is regulated and phosphorylated by the essential histidine kinase CckA and the histidine phosphotransferase ChpT. ChpT branches the CckA-ChpT-CtrA pathway to additionally control the proteolysis adapter protein CpdR negatively^153,154. CpdR first needs to be dephosphorylated to act as a polar localization factor and as an adaptor for ClpXP to degrade CtrA with the adaptor proteins RcdA and PopA^155,156. This happens when CckA switches from a kinase to a phosphatase during G1-to-S phase transition, dephosphorylating ChpT and subsequently CtrA and CpdR¹⁵⁷. This kinase to phosphatase switch is promoted by DivK which needs to be phosphorylated during swarmer to stalked cell transition.

In the swarmer cell, PleC sits at the flagellated pole and acts as a phosphatase on DivK, thereby keeping it inactive and cytoplasmic. Early in the G1-to-S phase transition, DivJ kinase levels begin to rise. DivJ phosphorylates DivK and DivK~P is then localized to the pole¹⁵⁸. This induces a positive feedback loop in which DivK~P switches PleC from a phosphatase to a kinase, which phosphorylates more DivK. Enforcing this feedback loop, DivK~P also enhances DivJ kinase activity further. PleC and DivJ together phosphorylate the diguanylate cyclase PleD, which leads to an increase of cellular c-di-GMP levels ²⁷.

To control the CckA-ChpT-CtrA pathway, DivK~P localizes to the pole and inhibits the noncanonical kinase DivL, a protein that keeps the CckA-ChpT-CtrA pathway active in the swarmer cell. DivL then localizes CckA to the pole¹⁵⁹. However, some CckA is still present in the cytoplasm. There, c-di-GMP acts as an additional switch on delocalized CckA molecules to tightly move the entire cellular pool into phosphatase mode¹⁰. All these processes together release the cell into S phase.

During the G1-to-S phase transition, DivJ is localized to the flagellated pole by SpmX and PopZ where the flagellum is shed, and the stalk and holdfast are produced^151,160. PleC is then localized to the opposite pole by the freshly expressed localization factor PodJL in the predivisional cell^161,162. There it likely sequesters DivK~P, allowing DivL to become active again and establishes the two cell fates after division¹⁶³ (Figure 1.7).

18 | Introduction

1.7.3 Timed protein degradation is essential for the Caulobacter crescentus cell cycle

In C. crescentus, a cascade of adapters guide proteins to ClpXP during the cell cycle. These adapters bind to each other in a hierarchical manner, and allow controlled degradation at specific time-points during G1-to-S phase transition¹⁵⁶. The first adaptor for ClpXP is the CckA- ChpT controlled protein CpdR. After the CckA kinase to phosphatase switch, this protein targets the phosphodiesterase PdeA for ClpXP degradation, allowing for an increase of c-di- GMP levels in the cell. This corresponds with the activation of the diguanylate cyclase PleD¹⁶⁴. In addition to binding to PdeA, the ClpXP-CpdR complex can also bind the next adaptor RcdA.

RcdA targets the stalk regulator TacA and other proteins for degradation^70,156. RcdA substrates compete with the last adaptor PopA. This means PopA can also act as an antiadaptor, inhibiting RcdA mediated degradation. To be active and degrade CtrA, PopA needs to bind c-di- GMP. CtrA degradation is therefore dependent on high c-di-GMP levels^104,156, meaning that protein degradation is closely linked to c-di-GMP control. This is also the case for the polar flagellum regulator TipF. TipF localizes to the pole upon binding to c-di-GMP where it recruits other factors involved in flagellar synthesis. When c-di-GMP levels are decreasing in the swarmer compartment of the predivisional cell, TipF is released from the pole and degraded by ClpXP¹⁶⁵.

Cleaving of the polar factor PodJ is a ClpXP independent example for complex proteolysis in C.

crescentus. Two forms of PodJ are involved in the cell cycle, PodJL and PodJS. PodJL localizes PleC to the swarmer pole. Then the periplasmic protease PerP truncates PodJ to PodJS with a new function in chemotaxis and holdfast formation. Finally yet another protease MmpA releases PodJS into the cytoplasm where it can be degraded¹⁶².

1.7.4 The ShkA-TacA pathway is necessary for G1-to-S-phase transition and development To initiate gene transcription, a class of proteins called sigma factors are needed. Two broad classes of sigma factors exist, the sigma 70 factor (σ⁷⁰) family and sigma 54 factor (σ⁵⁴) family¹⁶⁶.

Sigma factors guide the RNA polymerase specifically to the promoter and melt the DNA duplex to allow transcription. In this process, first the polymerase subunits assemble the core RNA polymerase complex. σ⁷⁰ binds and targets the protein complex to two conserved sequences, generally located -10 and -35 nucleotides upstream of the transcriptional start site, and melts the DNA in the -10 region¹⁶⁷.

The alternative σ⁵⁴ family is structurally and functionally distinct from the σ⁷⁰ family. In contrast to σ⁷⁰, σ⁵⁴ binds to a region -24 and -12 from the transcriptional start site with a highly conserved consensus¹⁶⁸. σ⁵⁴ expression can be temporally controlled. This is the case in C.

crescentus, where the σ⁵⁴ gene rpoN is expressed during the swarmer to stalked cell transition¹⁶⁹. However, there is an additional level of control. The most significant difference to σ⁷⁰ is given by the requirement of a bacterial enhancer binding protein (bEBP) to activate σ⁵⁴ transcription¹⁷⁰. bEBPs bind 80 to 150 bp upstream of the transcriptional starting site. In order to initiate melting of the DNA, bEBPs need to interact with the RNA polymerase complex and require energy from ATP¹⁷¹. According to the function of bEBPs, they consist usually of three domains: An AAA⁺ domain responsible for ATP hydrolysis, a domain containing a helix-turn- helix (HTH) motif that binds to the DNA, and a regulatory domain to control the protein¹⁷². The regulatory domain can be very diverse or entirely missing. In many cases however, bEBPs are part of a two-component system and regulated by phosphorylation from a histidine kinase¹⁷³. In C. crescentus, this is the case for the ShkA-TacA pathway. TacA is a transcription factor of the bEBP family¹⁷⁴. In this pathway, TacA is controlled by the hybrid histidine kinase ShkA that is as part of a His-Asp phosphorelay involved in cell morphogenesis and development.

Interestingly, this histidine kinase is predicted to lack any transmembrane or sensor domains and therefore belongs to a small subgroup of cytoplasmic regulators of which little is known.

Adding to the unusual protein architecture are two receiver domains that are attached to the kinase domain. The first one is a pseudo-receiver domain, lacking key residues that would allow it to be phosphorylated. The second receiver domain can be phosphorylated and transfers its phosphate to the phosphotransferase ShpA. ShpA then phosphorylates TacA (Figure 1.5).

The expression of tacA itself is dependent on CtrA and PleC¹⁵¹. The ShkA-TacA pathway is active early in the G1-to-S phase transition and quickly turned off by ClpXP degradation^70,174.

20 | Introduction

Figure 1.5: The ShkA-TacA pathway resembles a traditional phosphorelay. However, ShkA does not have a transmembrane domain or an input domain. The pseudoreceiver domain of ShkA cannot be phosphorylated, due to residues missing that are essential for M²⁺ binding.

Figure 1.6: Sequence analysis of the pseudo-receiver domain and the receiver domain of ShkA, reveals that the pseudo receiver domain cannot be phosphorylated (missing ion-binding site indicated in orange).

One of the genes controlled by the ShkA-TacA phosphorelay is staR, which is a transcription factor involved in C. crescentus stalk biogenesis and responsible for controlling stalk length¹⁷⁴. StaR is also involved in holdfast regulation by inhibiting the promoter of the hfiA gene. The HfiA protein inhibits HfsJ, a glycosyltransferase required for holdfast production¹⁷⁵. Holdfast production is additionally controlled by c-di-GMP, in a StaR-independent fashion¹⁷⁶.

The other well-described target gene of TacA is spmX. SpmX is a lysozyme-like protein that localizes the DivJ kinase to the future stalked cell pole in the swarmer cell and activates DivJ to phosphorylate the cell fate determinant DivK¹⁵¹. The regulatory protein SpmY might influence spmX expression and ShkA-TacA pathway activity¹⁷⁷. SpmX itself is localized to the pole by another factor, PopZ, by direct binding. At the pole SpmX forms oligomers of trimers¹⁶⁰. PopZ tethers the centromere in the swarmer cell to the flagellated pole by interacting with the chromosome partitioning protein ParB^178,179. During G1-to-S phase transition, PopZ switches function to act as a localization factor. Multiple proteins apart from SpmX need PopZ for

robust polar localization, including DivK, CpdR, RcdA, ClpX and CckA. PopZ remains its function at the stalked pole, however during replication, new PopZ localizes to the opposite pole regaining its centromere tethering function¹⁸⁰ (Figure 1.7).

Figure 1.7: Polar localization of key drivers of cell cycle progression in C. crescentus. The letter K indicates kinase mode, the letter P phosphatase mode. The peak of c-di-GMP levels is qualitatively shown as a grey curve. Low cellular c-di-GMP levels are indicated in blue, high cellular c-di-GMP levels in orange.

22 | Aim of the Thesis

2 A IM OF THE T HESIS

The second messenger c-di-GMP controls C. crescentus development and coordinates this process with cell cycle progression. Recently Lori et al. have shown, that c-di-GMP triggers S- phase entry and replication initiation by controlling the histidine kinase CckA¹⁰. However, the role of c-di-GMP in gene expression during the G1-S transition has remained elusive.

In a Capture Compound Mass Spectrometry (CCMS) screen, Jutta Nesper identified ShkA to be a putative c-di-GMP binder. Levels of c-di-GMP rise early in the cell cycle⁸⁷ during the G1-S transition when also the ShkA-TacA pathway is highly active.

The aim of my thesis was to investigate how c-di-GMP regulates the ShkA-TacA pathway to control cell cycle timing, gene expression and morphology during the G1-to-S phase transition.

3 P ROJECT A BSTRACT

The following chapters are based on the publication manuscript in preparation:

A c-di-GMP controlled phosphorelay drives a bacterial cell cycle by inducing a G1/S specific transcriptional network

Kaczmarczyk, A.*, Hempel, A.M.*, von Arx, C.*, Nesper, J., Jenal U.

* Equal contribution / order to be discussed

Bacteria can adapt their growth rates to the prevailing nutritional conditions. Proliferation rates are controlled by the cell cycle and are essentially determined by the step committing cells to initiate chromosome replication. The transcriptional network controlling this process is still largely unknown. Here we show that in the alpha-proteobacterium C. crescentus, a G1/S specific transcriptional network orchestrates entry into the replicative cycle. We demonstrate that fluctuating levels of the second messenger c-di-GMP control the ShkA-TacA phosphorylation cascade during the cell cycle, leading to the expression of a specific set of genes required to prepare cells for replication initiation and to coordinate this process with morphogenesis. C-di-GMP specifically binds to a pseudo-receiver domain of the ShkA kinase to alleviate its autoinhibitory function. This stimulates kinase activity, leads to the phosphorylation and activation of the TacA transcription factor and the activation of TacA- dependent gene expression during the G1/S transition. We further demonstrate that c-di-GMP mediated activation of ShkA together with delayed c-di-GMP dependent degradation of TacA, limiting gene expression to a narrow window during the G1/S transition. This is reminiscent of G1/S control in eukaryotes where cyclins and cyclin-dependent kinases define the timing of the G1/S specific transcriptional network by a combination of controlled phosphorylation and protein degradation.

24 | Project Introduction

4 P ROJECT I NTRODUCTION

Metazoan development and tissue homeostasis relies on self-renewal by quiescent stem cells that retain the capacity to re-enter the cell cycle. Conversely, misregulation of these cells results in uncontrolled proliferation and cancer¹⁸¹. A pivotal step of the cell cycle is the progression from G1 to S phase, which commits cells to chromosome replication and eventually to cytokinesis^182,183. The transcriptional network regulating cell cycle progression is ultimately controlled by members of the conserved cell cycle machinery including cyclin- dependent kinases and cyclins, their stage-specific regulatory subunits^184,185. Cell cycle control in bacteria is less well understood. Bacteria need to coordinate cell growth and differentiation with their metabolic status and nutrient availability. Their cell cycle is divided into three stages, the period between birth and the initiation of chromosome replication (B period or G1); the period required for chromosome replication (C period or S); and the time between the completion of replication and the end of cell division (D period or G2)¹⁸⁶. Since the C and D periods remain essentially constant over a wide range of growth rates¹⁸⁷, the step committing cells to initiate chromosome replication largely determines bacterial proliferation rates.

Moreover, chromosome replication initiates at a constant cell volume irrespective of the cell’s size at birth, indicating that this represents the initiating step of the bacterial cell cycle^188,189. Although recent work has improved our understanding of how cell growth and cell cycle progression are coupled (reviewed in^190,191), a molecular frame of the regulatory network driving G1/S transition is still largely missing.

To accomplish rapid growth, bacteria like E. coli or B. subtilis can initiate new rounds of replication before completing previous rounds of genome replication¹⁸⁷. In contrast, C.

crescentus strictly separates its cell cycle stages in a eukaryote-like manner, initiating chromosome replication only once per division cycle¹⁹². This, and the fact that different stages of the cell cycle have distinct morphologies and can be separated from each other experimentally, makes C. crescentus particularly amenable to study bacterial cell cycle control¹⁹³. Caulobacter divides asymmetrically to generate a sessile stalked cell (ST), which enters S-phase immediately and a motile swarmer cell (SW) that enters G1 phase, in which replication is blocked. After a defined period of motility, SW cells initiate a program to exit G1 into S-phase. Concomitant with initiating genome replication cells change morphology and behavior by replacing the rotary motor with an extension of the cell body, the stalk, and an adhesive holdfast that anchors sessile ST cells to surfaces. Replication initiation requires the